skip to main content
DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

This content will become publicly available on August 20, 2020

Title: Convolutional neural networks on the HEALPix sphere: a pixel-based algorithm and its application to CMB data analysis

Abstract

We describe a novel method for the application of convolutional neural networks (CNNs) to fields defined on the sphere, using the Hierarchical Equal Area Latitude Pixelization scheme (HEALPix). Specifically, we have developed a pixel-based approach to implement convolutional and pooling layers on the spherical surface, similarly to what is commonly done for CNNs applied to Euclidean space. The main advantage of our algorithm is to be fully integrable with existing, highly optimized libraries for NNs (e.g., PyTorch, TensorFlow, etc.). We present two applications of our method: (i) recognition of handwritten digits projected on the sphere; (ii) estimation of cosmological parameter from simulated maps of the cosmic microwave background (CMB). The latter represents the main target of this exploratory work, whose goal is to show the applicability of our CNN to CMB parameter estimation. We have built a simple NN architecture, consisting of four convolutional and pooling layers, and we have used it for all the applications explored herein. Concerning the recognition of handwritten digits, our CNN reaches an accuracy of ~95%, comparable with other existing spherical CNNs, and this is true regardless of the position and orientation of the image on the sphere. For CMB-related applications, we tested the CNNmore » on the estimation of a mock cosmological parameter, defining the angular scale at which the power spectrum of a Gaussian field projected on the sphere peaks. We estimated the value of this parameter directly from simulated maps, in several cases: temperature and polarization maps, presence of white noise, and partially covered maps. For temperature maps, the NN performances are comparable with those from standard spectrum-based Bayesian methods. For polarization, CNNs perform about a factor four worse than standard algorithms. Nonetheless, our results demonstrate, for the first time, that CNNs are able to extract information from polarization fields, both in full-sky and masked maps, and to distinguish between E and B-modes in pixel space. Lastly, we have applied our CNN to the estimation of the Thomson scattering optical depth at reionization ( τ) from simulated CMB maps. Even without any specific optimization of the NN architecture, we reach an accuracy comparable with standard Bayesian methods. This work represents a first step towards the exploitation of NNs in CMB parameter estimation and demonstrates the feasibility of our approach.« less

Authors:
 [1];  [2]
  1. International School for Advanced Studies, Trieste (Italy); Inst. for Fundamental Physics of the Univ., Trieste (Italy)
  2. Univ. degli Studi di Milano, Milano (Italy); Istituto Nazionale di Fisica Nucleare (INFN), Milano (Italy)
Publication Date:
Research Org.:
Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States). National Energy Research Scientific Computing Center (NERSC)
Sponsoring Org.:
USDOE Office of Science (SC)
OSTI Identifier:
1577584
Resource Type:
Accepted Manuscript
Journal Name:
Astronomy and Astrophysics
Additional Journal Information:
Journal Volume: 628; Journal ID: ISSN 0004-6361
Publisher:
EDP Sciences
Country of Publication:
United States
Language:
English
Subject:
79 ASTRONOMY AND ASTROPHYSICS; Astronomy & Astrophysics; Abstract; Full HTML; PDF (2.726 MB); ePUB (6.803 MB); References; NASA ADS Abstract Service

Citation Formats

Krachmalnicoff, N., and Tomasi, M. Convolutional neural networks on the HEALPix sphere: a pixel-based algorithm and its application to CMB data analysis. United States: N. p., 2019. Web. doi:10.1051/0004-6361/201935211.
Krachmalnicoff, N., & Tomasi, M. Convolutional neural networks on the HEALPix sphere: a pixel-based algorithm and its application to CMB data analysis. United States. doi:10.1051/0004-6361/201935211.
Krachmalnicoff, N., and Tomasi, M. Tue . "Convolutional neural networks on the HEALPix sphere: a pixel-based algorithm and its application to CMB data analysis". United States. doi:10.1051/0004-6361/201935211.
@article{osti_1577584,
title = {Convolutional neural networks on the HEALPix sphere: a pixel-based algorithm and its application to CMB data analysis},
author = {Krachmalnicoff, N. and Tomasi, M.},
abstractNote = {We describe a novel method for the application of convolutional neural networks (CNNs) to fields defined on the sphere, using the Hierarchical Equal Area Latitude Pixelization scheme (HEALPix). Specifically, we have developed a pixel-based approach to implement convolutional and pooling layers on the spherical surface, similarly to what is commonly done for CNNs applied to Euclidean space. The main advantage of our algorithm is to be fully integrable with existing, highly optimized libraries for NNs (e.g., PyTorch, TensorFlow, etc.). We present two applications of our method: (i) recognition of handwritten digits projected on the sphere; (ii) estimation of cosmological parameter from simulated maps of the cosmic microwave background (CMB). The latter represents the main target of this exploratory work, whose goal is to show the applicability of our CNN to CMB parameter estimation. We have built a simple NN architecture, consisting of four convolutional and pooling layers, and we have used it for all the applications explored herein. Concerning the recognition of handwritten digits, our CNN reaches an accuracy of ~95%, comparable with other existing spherical CNNs, and this is true regardless of the position and orientation of the image on the sphere. For CMB-related applications, we tested the CNN on the estimation of a mock cosmological parameter, defining the angular scale at which the power spectrum of a Gaussian field projected on the sphere peaks. We estimated the value of this parameter directly from simulated maps, in several cases: temperature and polarization maps, presence of white noise, and partially covered maps. For temperature maps, the NN performances are comparable with those from standard spectrum-based Bayesian methods. For polarization, CNNs perform about a factor four worse than standard algorithms. Nonetheless, our results demonstrate, for the first time, that CNNs are able to extract information from polarization fields, both in full-sky and masked maps, and to distinguish between E and B-modes in pixel space. Lastly, we have applied our CNN to the estimation of the Thomson scattering optical depth at reionization (τ) from simulated CMB maps. Even without any specific optimization of the NN architecture, we reach an accuracy comparable with standard Bayesian methods. This work represents a first step towards the exploitation of NNs in CMB parameter estimation and demonstrates the feasibility of our approach.},
doi = {10.1051/0004-6361/201935211},
journal = {Astronomy and Astrophysics},
number = ,
volume = 628,
place = {United States},
year = {2019},
month = {8}
}

Journal Article:
Free Publicly Available Full Text
This content will become publicly available on August 20, 2020
Publisher's Version of Record

Save / Share:

Works referenced in this record:

FERMI LARGE AREA TELESCOPE THIRD SOURCE CATALOG
journal, June 2015

  • Acero, F.; Ackermann, M.; Ajello, M.
  • The Astrophysical Journal Supplement Series, Vol. 218, Issue 2
  • DOI: 10.1088/0067-0049/218/2/23

The clustering of galaxies in the SDSS-III Baryon Oscillation Spectroscopic Survey: baryon acoustic oscillations in the Data Release 9 spectroscopic galaxy sample
journal, December 2012


Rotation-invariant convolutional neural networks for galaxy morphology prediction
journal, April 2015

  • Dieleman, Sander; Willett, Kyle W.; Dambre, Joni
  • Monthly Notices of the Royal Astronomical Society, Vol. 450, Issue 2
  • DOI: 10.1093/mnras/stv632

SkyNet: an efficient and robust neural network training tool for machine learning in astronomy
journal, May 2014

  • Graff, Philip; Feroz, Farhan; Hobson, Michael P.
  • Monthly Notices of the Royal Astronomical Society, Vol. 441, Issue 2
  • DOI: 10.1093/mnras/stu642

Enzo: an Adaptive mesh Refinement code for Astrophysics
journal, March 2014

  • Bryan, Greg L.; Norman, Michael L.; O'Shea, Brian W.
  • The Astrophysical Journal Supplement Series, Vol. 211, Issue 2
  • DOI: 10.1088/0067-0049/211/2/19

A TWO-DIMENSIONAL INFRARED MAP OF THE EXTRASOLAR PLANET HD 189733b
journal, February 2012


XSPECT, estimation of the angular power spectrum by computing cross-power spectra with analytical error bars
journal, April 2005


Cosmological parameters from CMB and other data: A Monte Carlo approach
journal, November 2002


Computational biology perspective: kernel methods and deep learning: Kernel methods and deep learning
journal, July 2012

  • Lodhi, Huma
  • Wiley Interdisciplinary Reviews: Computational Statistics, Vol. 4, Issue 5
  • DOI: 10.1002/wics.1223

DeepSphere: Efficient spherical convolutional neural network with HEALPix sampling for cosmological applications
journal, April 2019


Neural networks and the separation of cosmic microwave background and astrophysical signals in sky maps
journal, November 2000


ANN z : Estimating Photometric Redshifts Using Artificial Neural Networks
journal, April 2004

  • Collister, Adrian A.; Lahav, Ofer
  • Publications of the Astronomical Society of the Pacific, Vol. 116, Issue 818
  • DOI: 10.1086/383254

Model for Evaluation of Stock Values by Ensemble Model Using Deep Learning [深層学習を用いたアンサンブルモデルによる株主価値推定モデルの提案]
journal, January 2018

  • Tamura, Koichiro; Uenoyama, Katsuya; Iitsuka, Shuhei
  • Transactions of the Japanese Society for Artificial Intelligence, Vol. 33, Issue 1
  • DOI: 10.1527/tjsai.A-H51

Planck 2015 results : I. Overview of products and scientific results
journal, September 2016


Financial Time Series Prediction Based on Deep Learning
journal, December 2017


HEALPix: A Framework for High‐Resolution Discretization and Fast Analysis of Data Distributed on the Sphere
journal, April 2005

  • Gorski, K. M.; Hivon, E.; Banday, A. J.
  • The Astrophysical Journal, Vol. 622, Issue 2
  • DOI: 10.1086/427976

Deep learning for computational biology
journal, July 2016

  • Angermueller, Christof; Pärnamaa, Tanel; Parts, Leopold
  • Molecular Systems Biology, Vol. 12, Issue 7
  • DOI: 10.15252/msb.20156651

The Transiting Exoplanet Survey Satellite: Simulations of Planet Detections and Astrophysical False Positives
journal, August 2015

  • Sullivan, Peter W.; Winn, Joshua N.; Berta-Thompson, Zachory K.
  • The Astrophysical Journal, Vol. 809, Issue 1
  • DOI: 10.1088/0004-637X/809/1/77

Fast cosmological parameter estimation using neural networks
journal, March 2007


Polarized CMB power spectrum estimation using the pure pseudo-cross-spectrum approach
journal, June 2009


Measuring the Baryon Acoustic Oscillation scale using the Sloan Digital Sky Survey and 2dF Galaxy Redshift Survey: Measuring the BAO scale
journal, September 2007

  • Percival, Will J.; Cole, Shaun; Eisenstein, Daniel J.
  • Monthly Notices of the Royal Astronomical Society, Vol. 381, Issue 3
  • DOI: 10.1111/j.1365-2966.2007.12268.x

Rapid Bayesian position reconstruction for gravitational-wave transients
journal, January 2016


Neural Network Ensemble Based CAD System for Focal Liver Lesions from B-Mode Ultrasound
journal, April 2014


DeepCMB: Lensing reconstruction of the cosmic microwave background with deep neural networks
journal, July 2019


A hybrid approach based on neural networks and genetic algorithms for detecting temporal patterns in stock markets
journal, March 2007


Role of Artificial Intelligence Techniques (Automatic Classifiers) in Molecular Imaging Modalities in Neurodegenerative Diseases
journal, January 2017


Observation of High-Energy Astrophysical Neutrinos in Three Years of IceCube Data
journal, September 2014


Earth as an Extrasolar Planet: Earth Model Validation Using EPOXI Earth Observations
journal, June 2011

  • Robinson, Tyler D.; Meadows, Victoria S.; Crisp, David
  • Astrobiology, Vol. 11, Issue 5
  • DOI: 10.1089/ast.2011.0642